Freqtrade Alternative: Trigr vs Freqtrade and Hummingbot

TL;DR: Freqtrade and Hummingbot are mature, free, open-source Python bots that you install, configure and host yourself; Freqtrade focuses on signal-based strategies with backtesting and hyperparameter optimization, and Hummingbot on market making across a wide range of centralized and decentralized exchange connectors. Trigr is a hosted alternative for Hyperliquid perps: no-code or AI-built strategies, point-in-time backtests with explicit costs, and non-custodial agents without a server. If you want full code control and many exchanges, the open-source bots win; if you want a shorter path from idea to a monitored perp agent, Trigr is built for that.

What are Freqtrade and Hummingbot?

Both are well-known open-source projects with active communities, and both are written in Python. They solve different problems, which matters before comparing either with Trigr.

Freqtrade describes itself on its official site as a free and open-source crypto trading bot. Strategies are Python classes built on pandas. It includes backtesting on downloaded historical data, hyperparameter optimization ("hyperopt"), an optional machine-learning module called FreqAI, a dry-run mode with simulated money, and control through Telegram or a web UI. It supports spot on many exchanges and leveraged futures on several, including Hyperliquid.

Hummingbot describes itself as an open-source framework for crypto market makers, released under the Apache 2.0 license. Its strength is breadth of connectors across centralized and decentralized exchanges and strategies such as pure market making and arbitrage. Newer V2 "controllers" package strategies so they can be backtested and deployed through the Hummingbot Dashboard. It has a dedicated Hyperliquid connector for spot and perpetuals.

Features and supported exchanges change often in open-source projects, so treat the details here as accurate as of September 2026.

How is Trigr different in approach?

Trigr is not a framework you install. It is a hosted platform focused on one job: building, testing and running strategies on Hyperliquid perpetual futures, including TradFi perps via HIP-3 for backtests and paper agents in the current beta.

• Strategies are graphs, not code. A strategy is a node graph of TRIGGER, FILTER, SIGNAL, RISK and EXECUTE, built in the Studio, drafted by an AI copilot, or created from ChatGPT, Claude or Codex over the Trigr MCP server.
• Fixed, conservative backtest conventions. As the backtesting docs describe: point-in-time data, next-bar-open fills, stops winning when an intrabar tie cannot be resolved, and fees always applied. Slippage and funding are opt-in, and a first result is labeled gross.
• Nothing to host. Agents run on Trigr's infrastructure. On Hyperliquid they use a trade-only agent key that cannot withdraw.
• Alternative data built in. Funding rate, open interest, liquidations, long/short ratio, DVOL and macro series from FRED, the VIX, CFTC COT and the EIA are available as graph nodes.

Side-by-side comparison

How do the backtests differ?

Freqtrade is refreshingly explicit about its assumptions. Its backtesting documentation says entries happen at the open price unless custom pricing is used, exit-signal exits happen at the open of the next candle, and orders fill at the requested price with no slippage as long as that price is inside the candle's range. Fees default to the exchange's fee and can be overridden. The documentation also states plainly that backtesting will never replace a dry run.

Trigr's defaults overlap more than they differ: next-bar-open fills and fees always on, the same baseline covered in the comparison with TradingView's Strategy Tester. The differences are in the details:

• Slippage. Trigr lets you add a flat basis-points cost per fill against you. It is a simple assumption, not an order-book model, but it lets you see the net result next to the gross one.
• Funding. Freqtrade applies funding fees in futures mode and lets you set a fallback funding rate for periods without data. Trigr accrues funding at roughly 8-hour settlements, signed by side, from Binance's historical USDT-M series or a flat rate you choose. The post on slippage and funding in perp backtests explains why this changes slow strategies the most.
• Intrabar ties. When a take-profit and stop-loss are both touched in one bar, Trigr replays 5-minute bars to see which came first, and if it still cannot tell, the stop wins.
• History on Hyperliquid. Freqtrade's exchange notes say that Hyperliquid's API provides limited historic candles, so downloading deep Hyperliquid history is not possible there. Trigr's standard backtests use full available history, with coarser cross-asset inputs carried forward, never read ahead.

Optimization and overfitting

Freqtrade's hyperopt searches parameter combinations for the best result on historical data. It is a powerful tool, and like any search it raises a selection-bias question: the best of many runs is partly luck.

Trigr takes a narrower route. Its ML optimization chooses among LightGBM, random forest and XGBoost by anchored, purged walk-forward out-of-sample performance, with no hyperparameter tuning, and reports the Deflated Sharpe Ratio using the run's trial count plus the Probability of Backtest Overfitting. A plain backtest does not compute DSR or PBO; it labels them as not yet computed. The explainer on the Deflated Sharpe Ratio covers why a trial count matters.

What does running it involve?

This is the biggest practical difference.

With Freqtrade or Hummingbot you are the operator. You install Python or Docker, keep dependencies current, secure the machine that holds your keys, keep the process alive, and watch logs or Telegram messages. Both projects publish sensible key advice. Freqtrade's Hyperliquid notes recommend a separate API wallet rather than your main wallet's private key, and recommend never storing your main wallet key on the server.

With Trigr, the running part is managed:

• A Hyperliquid agent uses a trade-only agent key (an API wallet) that can place and cancel orders but cannot withdraw or send funds to another address. For autonomous agents, Trigr stores that key encrypted server-side; you keep the master wallet key and custody. The guide to Hyperliquid agent wallets walks through it.
• Every live entry is sent with reduce-only take-profit and stop-loss orders, and each agent is reconciled against Hyperliquid every minute.
• An agent can hold up to six strategies on different assets, with a combined backtest and a frontier marker between backtest and forward record.
• Paper agents let you forward-test first, filling at the current Hyperliquid price with the taker fee and builder fee applied.

The trade-off is control. You cannot patch Trigr's engine, add an exchange, or write arbitrary Python logic inside a strategy.

When are Freqtrade or Hummingbot the better choice?

There are clear cases where the open-source bots fit better:

• You want to write code. Freqtrade strategies can express anything pandas can. Trigr's graph has fixed building blocks: exactly one trigger, filters, a signal and a RISK block with no custom exit graphs.
• You trade many exchanges or spot markets. Both projects cover far more venues. Trigr trades Hyperliquid perps live, plus Propr challenge accounts and paper.
• You want market making or arbitrage. That is Hummingbot's core purpose. Trigr does not run market-making strategies.
• You want parameter search or custom ML. Hyperopt and FreqAI give you more knobs. Trigr deliberately does no hyperparameter tuning.
• You want zero platform cost and full ownership. Open-source software is free to run, apart from your server and your time, and you can audit every line.

When is Trigr the better choice?

• You trade Hyperliquid perps and want funding, open interest, liquidations and macro data without writing ingestion code.
• You would rather not operate a server, manage a Python environment, or keep a process alive.
• You want gross and net results side by side, conservative intrabar handling, and an audit trail of real versus simulated inputs.
• You want to build with an AI assistant under explicit approvals. Over MCP, an assistant can draft and backtest strategies and create a paused paper agent, but cannot start execution, place a live trade or withdraw.
• You want one path from idea to backtest to paper to a live non-custodial agent.

Trigr's published pricing includes a Free plan with 3,000 one-time trial credits, Trader at $29 per month and Pro at $99 per month; a standard backtest costs 50 credits and cache hits are free. Live trading is available on every plan, and the Hyperliquid builder fee is up to 0.05% per trade.

Backtests are not guarantees; perps are leveraged and can lose more than expected, whichever tool you choose.

Next steps

If you are coming from Freqtrade, rebuild one strategy's entry logic as a graph with the no-code strategy builder guide, run it gross and then net, and compare the trade logs side by side.